Tuning Stage Research Articles

LogFormer: A Pre-train and Tuning Pipeline for Log Anomaly Detection

Log anomaly detection is a key component in the field of artificial intelligence for IT operations (AIOps). Considering log data of variant domains, retraining the whole network for unknown domains is inefficient in real industrial scenarios. However, previous deep models merely focused on extracting the semantics of log sequences in the same domain, leading to poor generalization on multi-domain logs. To alleviate this issue, we propose a unified Transformer-based framework for Log anomaly detection (LogFormer) to improve the generalization ability across different domains, where we establish a two-stage process including the pre-training and adapter-based tuning stage. Specifically, our model is first pre-trained on the source domain to obtain shared semantic knowledge of log data. Then, we transfer such knowledge to the target domain via shared parameters. Besides, the Log-Attention module is proposed to supplement the information ignored by the log-paring. The proposed method is evaluated on three public datasets and one real-world dataset. Experimental results on multiple benchmarks demonstrate the effectiveness of our LogFormer with fewer trainable parameters and lower training costs.

Hybrid Active Noise Control with Auxiliary Filter-based Virtual Sensing

In this paper, we propose a hybrid active noise control system with auxiliary filter-based virtual sensing. The proposed system consists of a Feed-Forward (FF) part, which can reduce broadband noise, and a FeedBack (FB) part, which can reduce narrowband disturbance components, by introducing virtual sensing into the hybrid ANC. Virtual sensing is realized by using two auxiliary filters because the optimal noise control filters are different for the FF and FB parts, respectively. These auxiliary filters store information on the optimal noise control filters that can reduce both narrowband disturbance and broadband noise at the desired location in the tuning stage. Through simulations with real impulse responses for real applications, we demonstrate that the proposed hybrid ANC can achieve an overall noise reduction of about 30 dB.

Machine Learning Approach for New COVID-19 Cases Using Recurrent Neural Networks and Long-Short Term Memory

This research aims to predict new COVID-19 cases in Bandung, Indonesia. The system implemented two types of deep learning methods to predict this. They were the recurrent neural networks (RNN) and long-short-term memory (LSTM) algorithms. The data used in this study were the numbers of confirmed COVID-19 cases in Bandung from March 2020 to December 2020. Pre-processing of the data was carried out, namely data splitting and scaling, to get optimal results. During model training, the hyperparameter tuning stage was carried out on the sequence length and the number of layers. The results showed that RNN gave a better performance. The test used the RMSE, MAE, and R2 evaluation methods, with the best numbers being 0.66975075, 0.47075, 0.29616625, and 0.7644 on the test data.

Innovation and Practice of Educational Management System in Colleges and Universities Based on the Concept of Data Analysis Education

Abstract This paper proposes the evaluation method of college education management mode based on deep belief network (DBN), establishes the evaluation index of college education management mode, takes it as the input of the deep belief network, adopts the divergence algorithm and backpropagation algorithm to learn and optimize the parameters of the deep belief network, and establishes the evaluation model of the education management mode of universities. The evaluation method of the college education management model evaluation of the deep belief network is verified by evaluating the results and performing error analysis on 438 questionnaire test samples. The results show that the number of training steps of the adopted DBN model in the supervised tuning stage is 214, and the error fluctuation range is ±0.6, which achieves the expected effect, indicating that the adoption of the model has accuracy in the evaluation of college education management model. Under the concept of data analysis education, the evaluation method proposed in this paper has a positive significance for the innovation and practice of educational and teaching management systems.

Applied Cavity Perturbation Used for Frequency Identification of Channel Ports in Manifold-Coupled Multiplexers

We investigate a practical technique for deembedding the channel filter S-parameters of manifold-coupled multiplexers (MUXs), without detaching filters from the manifold. The method is applicable for MUXs with an arbitrary number of channels and can be used for the device regardless of its bandwidth, guard band, or loss of filters. We reconfigure the N-port MUX to two simpler networks cascaded to each other. We assume that the manifold response is unknown and use the idea of applied perturbation on the channel filter, and then by comparing the response of the overall cascaded network, before and after the perturbation, we approximate the channel port response. The technique is useful in fast detecting the unexpectedly detuned channels or likely faults in the device without unnecessary plugging/unplugging; it is also useful in roughly tuning of the channel filers at the early stage of MUX tuning. The technique can be easily traced by the telecommunication community.

Online auto-tuning of multiresonant current controller with nature-inspired optimization algorithms and disturbance in the loop approach

This paper presents a novel tuning method of a complex multiresonant current controller (MRCC). An online auto-tuning process is proposed to provide optimal coefficients suitable for the real-time operation of a grid-connected power converter. The time-consuming and inaccurate simulation stage based on the model of the plant is omitted. A novel function of constraints has been introduced into the optimization scheme to ensure the desired behavior of the control system. In this solution, the noise level of a control signal and the rise time of controlled currents are directly defined. To the author’s best knowledge, it is the first time when such a solution is proposed. Next, the most commonly used meta-heuristic optimization algorithms were applied to solve this particular optimization problem. Another original concept presented in the paper is the application of the disturbance-in-the-loop (DiL) approach. In this method, the distorted phase voltages are emulated during the tuning stage, and optimal coefficients of MRCC are selected in the off-grid operation of the power converter. The proposed solution ensures safe and efficient online tuning of the MRCC for an extremely simple performance index. As a result, the high-performance operation of the grid-connected inverter is achieved. A brief stability analysis based on Lyapunov’s stability theory is included. Experimental results obtained for grid-connected inverter indicate superior disturbance attenuation and current tracking. Advantages of the proposed approach are exhibited when compared to two analytical reference solutions.

Representation Learning-Driven Fully Automated Framework for the Inverse Design of Frequency-Selective Surfaces

Frequency-selective surfaces (FSSs) refer to planar structures that behave with specific electromagnetic (EM) responses within a frequency range and are widely applied in wireless propagation systems. Given the fact that different EM responses correspond to distinguished topologies, conventional inverse design methods of FSSs are usually labor-intensive, as they rely on experienced human engineers to determine the topology and then rationally tune its structures. There have been great attempts using optimization algorithms (e.g., genetic algorithms) or machine learning to automate the second tuning stage after the initial EM topologies are determined by human engineers. However, the first topology selection stage still requires engagements with experienced engineers. This article proposes a fully automated framework for the inverse design of FSSs. We achieved a fully automated inverse design by establishing a machine-friendly mapping flow. The mapping flow derives its continuity and compactness from representation learning, which enables both autoselection of the topology and autoevolution of the unit cell based on the topology. The autoselection stage automatically determines the appropriate topology by compressing the EM constraints through the principal component analysis (PCA) and classifying the topology using the support vector machine (SVM). Afterward, the autoevolution system can efficiently evolve until it yields an optimal unit cell. We developed a self-monitor strategy to control the evolution and maximize the evolution efficiency by adaptively tuning the three modules within the autoevolution system. We validated the presented framework with four FSS designs. The results proved its potential as a highly efficient fully automated tool for the inverse design of FSSs.

Swarm Intelligence-Homotopy Hybrid Optimization-Based ANN Model for Tunable Bandpass Filter

High-performance tunable radio frequency (RF)/ microwave and millimeter-wave filter design is a challenging task due to the lack of a basic theory. The filtering characteristics are highly sensitive to the variation of tuning elements that are commonly modeled and achieved by optimization algorithms. However, those optimizations only provide satisfactory results with a good set of initial parameters. Such range-limited optimization algorithms generally have issues of falling into local optima, slow convergence, and cumbersome implementation. To mitigate this problem, for the first time, a topology-based local optimizer is integrated with metaheuristic global optimization algorithms in this work. We have hybridized the homotopy method with an improved whale optimization algorithm (WOA) and a gray wolf optimization (GWO) algorithm. In this work, an artificial neural network (ANN) is formulated and studied, which has twofold applications. First, ANN is used as a surrogate model to represent the time-consuming electromagnetic (EM) model in expediting the hybrid optimization process of tunable filters. Second, an ANN model is developed on data generated by the proposed optimization algorithm for predicting tunable circuit parameters at different tuning stages. The proposed ANN model-based algorithm is then applied to a fifth-order lumped-element tunable circuit and two fourth-order full-wave EM simulation models of two tunable bandpass filters (tBPFs). The calculated results out of the ANN model demonstrate a good agreement with simulation and measurement counterparts.

Multi-Stage Prompt Tuning for Political Perspective Detection in Low-Resource Settings

Political perspective detection in news media—identifying political bias in news articles—is an essential but challenging low-resource task. Prompt-based learning (i.e., discrete prompting and prompt tuning) achieves promising results in low-resource scenarios by adapting a pre-trained model to handle new tasks. However, these approaches suffer performance degradation when the target task involves a textual domain (e.g., a political domain) different from the pre-training task (e.g., masked language modeling on a general corpus). In this paper, we develop a novel multi-stage prompt tuning framework for political perspective detection. Our method involves two sequential stages: a domain- and task-specific prompt tuning stage. In the first stage, we tune the domain-specific prompts based on a masked political phrase prediction (MP3) task to adjust the language model to the political domain. In the second task-specific prompt tuning stage, we only tune task-specific prompts with a frozen language model and domain-specific prompts for downstream tasks. The experimental results demonstrate that our method significantly outperforms fine-tuning (i.e., model tuning) methods and state-of-the-art prompt tuning methods on the SemEval-2019 Task 4: Hyperpartisan News Detection and AllSides datasets.

SARS-CoV-2 Detection from Lung CT-Scan Images Using Fine Tuning Concept on Deep-CNN Pretrained Model

The problem of the spread of COVID-19 (SARS-CoV-2) is spreading fleetly and worldwide. Beforehand discovery and opinion of complaint is veritably important to ensure the right remedy so that it needs to be enforced through various practical approaches. In former studies, complaint discovery through medical imaging has started to appear and get a good delicacy of around 80 to 90 percent using machine learning. In the deep learning era, some trials get better accuracy of 95 percent using the traditional deep learning approach. Now, deep learning has developed more fleetly, especially for image classification. therefore, it's necessary to experiment with a pretrained model approach to medical images. In addition, the fine tuning approach will also be an aspect of the approach that will be carried out in this trial to be compared and to find out its effect, specifically on CT-Scan images of the lungs for the discovery of COVID 19. The results of this experiment showed that the pretrained model approach can get high accuracy. Relatively high accuracy, the smallest testing accuracy in this trial reached 94.78 percent of the Xception without fine tuning phase, this result has beaten the machine learning approach which is didn't reach 90 percent of accuracy. The best experiment testing accuracy get 97.59 percet on the VGG 16 by applying fine tuning. The results of this trial also show that the fine tuning stage (for the top 10th layers) can increase the accuracy of the model.

Performance of a Wall-Following Robot Controlled by a PID-BA using Bat Algorithm Approach

A wall-following robot needs a controller that applies the closed-loop concept to move actively without hindrance. Some controllers with good capabilities can act as controllers for wall follower robots, such as PID controllers. Conceptually, this controller's good performance depends on tuning the three gains before use. Instead of giving the expected and appropriate output, wrong settings will provide inaccuracies for the controller, so applying the manual method at the tuning stage is not recommended. For this reason, PID controllers are often implemented in a system supported by appropriate optimization methods, such as Genetic Algorithm or Particle Swarm Optimization. Furthermore, different from this, in this study, the Bath Algorithm is used as an alternative optimization algorithm. Its application begins with a realistic simulation of a wall-following robot. This is done to provide the possibility to implement online PID controllers and BAs. In the end, several methods are compared to find out the performance of this type of approach. Moreover, based on the observed comparative results, the proposed method gives a better value in accumulative error and convergence speed in the PID optimization process.

TWO-STAGE METHOD FOR PARAMETER ESTIMATING IN UNKNOWN NOISE ENVIRONMENT

This paper addresses the parameter estimation problem in the case of an unknown noise environment. A two-stage method consisting of the tuning and estimating stages is pro-posed for solving this problem. At the tuning stage, the estimator is tuned to the noise envi-ronment by minimizing the estimation root-mean-square error for a known data fragment in the space of the three free parameters; these free parameters allow modifying the behavior of the minimization problem objective function. At the estimating stage, estimation is car-ried out by solving the corresponding minimization problem using already tuned free pa-rameters. The features of the proposed method, including algorithms for the tuning and es-timating stages, are demonstrated for estimating a Gaussian pulse that slowly moves in the unknown noise environment and locates on a known constant background. The numerical simulations confirm the high performance of the proposed method.

An Algorithm for Estimation of Stator Resistance and Inductance of Low-Cost SMPMSM Drives

The information on stator resistance and inductance is extremely important for advanced electrical drives, which are used for tuning of position estimators, current controllers, temperature monitoring, and so on. The properly tuned control systems can provide fast response, have higher stability, and increase the efficiency of motor drives; as a result, it is of great importance at the development and tuning stages. This article proposes a simple estimation technique for surface-mounted permanent magnet synchronous motors (SMPMSMs), capable of measuring stator resistance and inductance. In order to measure the stator resistance, the developed algorithm injects two-leveled dc currents that provide information for resistance calculation. At the next stage, the estimation algorithm initiates exponential current transient, measures electric time constant of the motor, and calculates inductance. In contradiction to the conventional approach, this work composes a more detailed model of the motor and takes inverter voltage drop into account. Furthermore, this article studies several exponential transients, compares them to each other, and provides recommendations for use of each approach. Then, the proposed ideas are verified using several motors of different powers and parameters. Finally, the recommendations on application of each approach and parameter selection are provided.

Hyperparameter tuning for multi-label classification of feedbacks in online courses

In this work, we propose the extension of a methodology for the multi-label classification of feedback according to the Hattie and Timperley feedback model, incorporating a hyperparameter tuning stage. It is analyzed whether the incorporation of the hyperparameter tuning stage prior to the execution of the algorithms support vector machines, random forest and multi-label k-nearest neighbors, improves the performance metrics of multi-label classifiers that automatically locate the feedback generated by a teacher to the activities sent by students in online courses on the Blackboard platform at the task, process, regulation, praise and other levels proposed in the feedback model by Hattie and Timperley. The grid search strategy is used to refine the hyperparameters of each algorithm. The results show that the adjustment of the hyperparameters improves the performance metrics for the data set used.

ADVERT: An Adaptive and Data-Driven Attention Enhancement Mechanism for Phishing Prevention

Attacks exploiting the innate and the acquired vulnerabilities of human users have posed severe threats to cybersecurity. This work proposes ADVERT, a human-technical solution that generates adaptive visual aids in real-time to prevent users from inadvertence and reduce their susceptibility to phishing attacks. Based on the eye-tracking data, we extract visual states and attention states as system-level sufficient statistics to characterize the user's visual behaviors and attention status. By adopting a data-driven approach and two learning feedback of different time scales, this work lays out a theoretical foundation to analyze, evaluate, and particularly modify humans' attention processes while they vet and recognize phishing emails. We corroborate the effectiveness, efficiency, and robustness of ADVERT through a case study based on the data set collected from human subject experiments conducted at New York University. The results show that the visual aids can statistically increase the attention level and improve the accuracy of phishing recognition from 74.6% to a minimum of 86%. The meta-adaptation can further improve the accuracy to 91.5% (resp. 93.7%) in less than 3 (resp. 50) tuning stages.

Эвристическая методика настройки скрытых марковских моделей для распознавания образов стохастических процессов

Предложена эвристическая методика настройки параметров скрытых марковских моделей для реализации в алгоритмах распознавания образов стохастических процессов, регистрируемых в форме последовательностей наблюдений. Методика позволяет получить рабочие модели по небольшому числу обучающих реализаций и включает выполнение двух этапов: предварительная настройка начальных параметров модели с использованием априорных данных и обучение с применением алгоритма Баума - Уэлча. На обоих этапах используется дополнительная процедура корректировки параметров модели, позволяющая устранять проблемы численного характера при их реализации в алгоритмах распознавания. Представлены результаты экспериментальной апробации методики на примере построения моделей для алгоритма распознавания образов трех стохастических процессов.

Embedding Model-Based Fast Meta Learning for Downlink Beamforming Adaptation

This paper studies the fast adaptive beamforming for the multiuser multiple-input single-output downlink. Existing deep learning-based approaches assume that training and testing channels follow the same distribution which causes task mismatch, when the testing environment changes. Although meta learning can deal with the task mismatch, it relies on labelled data and incurs high complexity in the pre-training and fine tuning stages. We propose a simple yet effective adaptive framework to solve the mismatch issue, which trains an embedding model as a transferable feature extractor, followed by fitting the support vector regression. Compared to the existing meta learning algorithm, our method does not necessarily need labelled data in the pre-training and does not need fine-tuning of the pre-trained model in the adaptation. The effectiveness of the proposed method is verified through two well-known applications, i.e., the signal to interference plus noise ratio balancing problem and the sum rate maximization problem. Furthermore, we extend our proposed method to online scenarios in non-stationary environments. Simulation results demonstrate the advantages of the proposed algorithm in terms of both performance and complexity. The proposed framework can also be applied to general radio resource management problems.

Detection and classification of network attacks using the deepneural network cascade

This article shows the relevance of developing a cascade of deep neural networks for detecting and classifying network attacks based on an analysis of the practical use of network intrusion detection systems to protect local computer networks. A cascade of deep neural networks consists of two elements. The first network is a hybrid deep neural network that contains convolutional neural network layers and long short-term memory layers to detect attacks. The second network is a CNN convolutional neural network for classifying the most popular classes of network attacks such as Fuzzers, Analysis, Backdoors, DoS, Exploits, Generic, Reconnais-sance, Shellcode, and Worms. At the stage of tuning and training the cascade of deep neural networks, the selection of hyperparame-ters was carried out, which made it possible to improve the quality of the model. Among the available public datasets, one ofthe current UNSW-NB15 datasets was selected, taking into account modern traffic. For the data set under consideration, a data prepro-cessing technology has been developed. The cascade of deep neural networks was trained, tested, and validated on the UNSW-NB15 dataset. The cascade of deep neural networks was tested on real network traffic, which showed its ability to detect and classify at-tacks in a computer network. The use of a cascade of deep neural networks, consisting of a hybrid neural network CNN + LSTM and a neural network CNNhas improved the accuracy of detecting and classifying attacks in computer networks and reduced the fre-quency of false alarms in detecting network attacks.

Degradation performance analysis and operation optimization of rotary-traveling wave oscillators for RF-CMOS applications

In this work, the impact of the parasitic effects caused by discontinuities on Resonant Rotary Traveling Wave Oscillators (RTWO) for RF-CMOS applications, is analyzed and modeled. By using experimental data in addition to the full wave representation, semi-empirical models are obtained for different discontinuities contained in an RTWO. Then, using the proposed models and 180 nm CMOS process parameters, an RTWO was designed with an operating frequency of 20 GHz, whilst the impact of the discontinuities was quantified in its output signal. The results show that the impact of the parasitic elements due to these discontinuities present a filtering effect leading to a distortion of the magnitude and number of harmonics in the output signal, with a reduction of 13.8% when the effects of the geometrical discontinuities were included. In order to mitigate the parasitic effects due to discontinuities, the use of topological (layout) methods in stack paths and the Möbius connection are proposed. Additionally, a process variation analysis was conducted and the results show that the operation frequency could vary in a range from 17.21 GHz to 18.91 GHz; while the output voltage could fluctuate between 1.17 V and 1.91 V depending on the process, bias voltage and temperature (PVT) conditions. These effects must be taken into account in the design of RTWOs to avoid inefficient tuning stages, especially for high frequency applications, and for advanced technologies where the impact of the analyzed discontinuities is increased.

Auto-Tuning of DC-DC Buck Converters Through the Modified Relay Feedback Test

This paper presents the design and application of the modified relay feedback test (MRFT) auto-tuning method to the PID voltage-mode control of dc-dc buck converters. The proposed method, which also includes a set of tuning rules, does not require knowledge of the converter or load parameters, and produces a controller with near-optimal dynamic performance for virtually any dc-dc buck converter. The proposed MRFT auto-tuning also guarantees a specified phase margin. An important feature of the MRFT auto-tuning is that it is non-complex and thus practical, consisting of a simple test stage and a quick tuning stage. In the test stage oscillations are excited using the MRFT algorithm, and the frequency and amplitude of these oscillations are measured. In the tuning stage, these measurements are used with the pre-designed tuning rules to calculate PID controller parameters. The designed MRFT auto-tuning is evaluated experimentally by using it to auto-tune four different dc-dc buck converter designs, where in each case parameters of the converter are unknown to the auto-tuning algorithm. The experimental results show good performance of the MRFT auto-tuned controller in all cases. A mathematical proof that the MRFT auto-tuning guarantees a specified phase margin is also provided.